Ethical AI for Inclusive Financial Services

Artificial Intelligence (AI) systems are becoming more and more ubiquitous in the financial industry. The increase in development and deployment is not without good reason: AI can be of use for all aspects of financial services from fraud detection to credit approval. However, despite its potential to revolutionize the industry it has been pointed out that the use of AI is not without risks. 

The functioning and outcomes of AI systems can become opaque, turning AI systems into black-boxes, with virtually no possibilities to audit or control them. Furthermore, historical bias in datasets or inadequately calibrated models might systematically disadvantage subgroups in society thereby perpetuating or even reinforcing existing disparities. To address these risks various ethical frameworks and guidelines for the responsible use of AI have been proposed by the industry, governments and international regulatory bodies. One of the most prominent among these are the EU Guidelines for Trustworthy AI.

Since their publication in 2019, however, limited research has been done on how these guidelines could formalize into regulations and how they relate to the current regulations for finance in the Netherlands. To explore these questions the Ethical AI for Inclusive Financial Services research project examined how the existing regulations in finance could contribute to the desired ‘human centered AI’ and to what extent new regulation might be necessary to achieve this goal. With De Volksbank as project and valorization partner research was done on how the ethical guidelines relate to the existing duty of care and explainability obligations banks currently hold when it comes to credit approval. The results indicate that some ethical aspects might be covered within the current regulations but work still needs to be done to ensure a ‘fit’ between the ethical principles endorsed by the European Union and the legal frameworks currently in place.

The report offers a first starting point to begin exploring the questions of ethical and legal AI applications in finance. It by no means intends to provide in an exhaustive analysis of these topics but, as an exploration, already gives a strong recommendation to financial industries that it would be in their best interest to start incorporating the ethical principles of AI in their AI development processes as with this new powers come new responsibilities.

Download the Research Report (in Dutch)

PhD student
Professor
More information

The Ethical AI for Inclusive Financial Services research project was conducted by ESL Prof. Evert Stamhuis, ESPhil PhD candidate Joris Krijger and Financial Law master students Jan-Willem van Doorn and Maarten Weekenborg with financial support from the Dynamics of Inclusive Prosperity Initiative.